Behavior of protonated cyclopropyl intermediates during polyalphaolefin synthesis: Mechanism and predicted product distribution

Article abstract of DOI:10.1002/poc.3059

A new mechanism for the origin of multiple skeletal isomers observed in the cationic dimerization of 1-decene is proposed, and products that should form based on this mechanism are predicted. A protonated cyclopropyl intermediate appeared to form directly from combination of 2-decyl carbocation with 1-decene; formation of this intermediate did not appear to occur via ring closure of a branched secondary carbocation. The authors propose that rapid, repeated isomerizations of the protonated cyclopropyl intermediates lead to multiple skeletal isomers in decene dimers. The proposed mechanism can account for structures previously identified in mixtures of decene dimers and butene dimers. Copyright

Full text of DOI:10.1002/poc.3059

Research Article
Received: 21 August 2012,
Revised: 7 September 2012,
Accepted: 12 September 2012,
Published online in Wiley Online Library: 4 October 2012
(wileyonlinelibrary.com) DOI: 10.1002/poc.3059
Behavior of protonated cyclopropyl intermediates
during polyalphaolefin synthesis: Mechanism
and predicted product distribution
Jeffrey C. Geea*, Brooke L. Smalla and Kenneth D. Hopea
A new mechanism for the origin of multiple skeletal isomers observed in the cationic dimerization of 1-decene is
proposed, and products that should form based on this mechanism are predicted. A protonated cyclopropyl
intermediate appeared to form directly from combination of 2-decyl carbocation with 1-decene; formation of this
intermediate did not appear to occur via ring closure of a branched secondary carbocation. The authors propose that
rapid, repeated isomerizations of the protonated cyclopropyl intermediates lead to multiple skeletal isomers in
decene dimers. The proposed mechanism can account for structures previously identified in mixtures of decene
dimers and butene dimers. Copyright © 2012 John Wiley & Sons, Ltd.
Keywords: carbenium ion; carbocation rearrangement; cationic oligomerization; protonated cyclopropyl intermediate;
polyalphaolefin; skeletal isomerization mechanism
has occurred that is extremely rapid relative to both proton
elimination and reaction with monomer (to make trimer).
INTRODUCTION
Previous investigators have suggested that rapid double bond
isomerization occurs in the monomer and that reaction of
internal carbenium ions with 1-decene leads to the formation
of multiple isomers in the C20 fraction.[3] Our observations do
not support this conclusion.
The mechanism of alpha olefin oligomerization to make polyal-
phaolefins (PAO) has been a subject of much investigation for
some time.[1–4] Of chief interest is an explanation for the origin
of multiple skeletal isomers. Nuclear magnetic resonance (NMR)
evidence has shown that most of the oligomers contain one
more methyl group than a simple mechanism would predict,
and Shubkin and coworkers proposed a protonated cyclopropyl
intermediate at the dimerization stage to account for skeletal
isomerization that could produce an additional methyl group.[1,4]
Other workers have also proposed that double bond isomeriza-
tion in the 1-decene monomer could account for structures
present in the products.[2,3] Indeed, detailed NMR[5] and gas
chromatography/mass spectrometry (GC/MS)[6] work have pro-
vided much information about the structures of products in
the saturated dimer fraction (C20 isomers or “PAO 2”) isolated
from 1-decene oligomerization.
A typical PAO manufacturing process involves the use of BF3
or other Lewis acid, in combination with a protic source
such as an alcohol or water, to induce cationic oligomerization of
1-decene or other linear alpha olefin.[7,8] These conditions produce
a mixture of branched oligomers with a distribution that peaks
around trimer. The hydrogenated products, once fractionated into
different viscosity/molecular weight fractions, are outstanding
lubricants. Their structures impart excellent viscosity indices and
low pour points, as well as outstanding oxidative stabilities.
The simple mechanism for cationic oligomerization[9] (Fig. 1)
does not explain the high number of isomers observed in these
products. Chromatographic analysis of the hydrogenated dimer
fraction shows well over 50 different isomers.[2,6] NMR analyses
on dimer and other fractions have shown that all of the skeletal
rearrangement may occur at the dimerization stage,[1] and much
analytical work has focused on the dimer fraction.
EXPERIMENTAL
Oligomerization reactions. In a typical oligomerization reaction, 5.0g of
olefin, 0.3g of n-nonane internal standard, and 0.050–0.12 g of AlCl3 or
AlBr3 were combined in a 20ml glass bottle with magnetic stir bar.
Sometimes, 1-butanol (0.3 mole 1-butanol per mole AlX3) was included,
but the 40–100ppm water in the olefin samples was enough to generate
an active catalyst. The bottle was placed in a heated metal block inside a
glove box having dry nitrogen atmosphere. Mixtures stirred at 300 rpm
and 25ꢀC or 90 ꢀC. Reaction mixtures were periodically sampled by
withdrawing a few drops of liquid using a glass pipette and adding them
to 0.5ml of cyclohexane solvent. Outside the glove box, mixtures were
quenched with
a few drops of isopropanol and analyzed by gas
chromatography. From the chromatographic data, percent conversion of
monomer to oligomer was calculated, as well as oligomer carbon number
distributions. Skeletal isomerization in monomer was detectable, and the
number and width of peaks for dimer and heavier products was a qualitative
measure of the variety of isomers present. The 1-decene was from Chevron
Phillips Chemical Company (The Woodlands, TX USA). Aluminum chloride
and aluminum bromide were from Sigma-Aldrich (St. Louis, MO USA).
Gas chromatographic analyses. The method was split injection on a gas
chromatograph with a flame ionization detector (FID). Initial oven temper-
ature was 70 ꢀC and increased 5 ꢀC/min to 130 ꢀC, then 20 ꢀC/min to
* Correspondence to: J. C. Gee, Chevron Phillips Chemical Company, 1862
Kingwood Drive, Kingwood, Texas.
E-mail: geejc@cpchem.com
The high number of isomers observed in the saturated dimer
fraction indicates that some type of carbenium ion isomerization
a J.C. Gee, B.L. Small, K.D. Hope
Chevron Phillips Chemical Company, 1862 Kingwood Drive, Kingwood, Texas
J. Phys. Org. Chem. 2012, 25 1409–1417
Copyright © 2012 John Wiley & Sons, Ltd.

J. C. GEE, B. L. SMALL AND K. D. HOPE
Figure 1. Simple mechanism accepted for cationic polymerization
300 ꢀC for 20 min. The column was an Alltech FSOT capillary column, 30 m
 0.25 mm  0.25 mm. Data analyses were performed using Chemstation
W
software. This method was appropriate for determining overall
conversion to oligomer, for tracking any potential isomerization in
monomer, and for confirming broad isomer distributions in dimer.
Synthesis of 2-octyl-1-decene. In a round-bottomed flask, 105.5 g of
tri-n-octyl aluminum, 1364.3 g of 1-octene, and 733.1 g of 1-decene were
combined. The mixture was heated and stirred under nitrogen at 120 ꢀC
for 9.7 days, then was quenched at 80 ꢀC with 500 ml of aqueous NaOH
(25 wt %). By GC/FID, the organic phase was 10.8 wt % octenes, 6.2 wt
% decenes, 39.2 wt % hexadecenes, 35.6 wt % octadecenes, and 8.3 wt
% eicocenes; the C16 and C18 fractions were about 95% vinylidene :
5% internals. The 2-octyl-1-decene was isolated by vacuum distillation.
Synthesis of 7-methyl-4/5-undecenes. This mixture was prepared by
dimerization of 1-hexene (Chevron Phillips Chemical Company, The
Woodlands, TX USA), according to procedures described previously.[10]
Hydrogenation of the product mixture[11] showed the C12 paraffinic
product to be 65% 5-methyl undecane and 30% n-dodecane. Double
bond placement[12] in the dodecenes showed that 45% of the double
bonds were between C4 and C5, while 55% were between C5 and
C6. The mixture was therefore about 29 wt % 7-methyl-4-undecenes,
36 wt % 7-methyl-5-undecenes, 15 wt % 4-dodecenes, and 15 wt %
5-dodecenes, with 5% other C12 isomers.
Figure 2. Gas chromatograms showing peaks for C20 olefin isomers at
47% and 75% conversion during 1-decene oligomerization
decene was still 1-decene; at 75% conversion, 65–70% of the
unconverted decene was still 1-decene. If rapid double bond
isomerization in the decene fraction is necessary to form many
different C20 isomers, then we would expect to see substantial
double bond isomerization in the decene fraction early in the
oligomerization process, but we do not.
Table 1 shows the distribution of olefin isomers in the decene
fraction recovered from a reaction mixture having 75% decene
conversion to oligomer. We used a previously reported analytical
method[12] to determine this distribution. The double bond is
clearly moving in a stepwise manner toward the internal posi-
tions, as 2-decene makes up almost half of the internal olefin.
Table 1 also shows data for the distribution of decene isomers
recovered from a typical BF3-catalyzed oligomerization. We
draw similar conclusions from both samples: Double bond
isomerization in the decene fraction is clearly slow, relative to
oligomerization, and skeletal isomerization has not occurred in
the monomer fraction.
If 1-decene isomerized to internal olefins that oligomerized
faster than 1-decene, we might not detect the internal decenes.
However, we have observed that a mixture of internal decenes
oligomerizes at a much slower rate than 1-decene, in agreement
with previous reports.[4] We isomerized 1-decene to a mixture of
linear internal decenes on a silicoaluminophosphate (SAPO)
catalyst.[11] This mixture required 18 h to achieve 44% conversion
to oligomer, while 1-decene achieved 72% conversion in about
half an hour, under the same oligomerization conditions. Fur-
thermore, in a competitive oligomerization with 1-tetradecene,
internal decenes were less reactive than the alpha olefin. We
began with a mixture of 50 mole % internal decenes and 50
Computer model. Original source code for the computer model was
W
written using the Delphi
XE WindowsW development environment
from Embarcadero (San Francisco, CA USA).
RESULTS AND DISCUSSION
We conducted a number of small-scale reactions using AlCl3 or
AlBr3, along with trace water or 1-butanol, to oligomerize
several different olefin monomers. Mixtures of 1-decene dimers
appeared to contain the same products, regardless of whether
we used aluminum halide or BF3 as the dimerization catalyst.
We chose solid aluminum halides for our work, because solid
catalysts were easier to use than gaseous BF3 on a small scale.
(We did not have to regulate pressure.) NMR data indicated that
decene dimers made using aluminum halide had 3.7 methyl
groups per molecule, in close agreement with 3.8 methyl groups
[1–5]
reported for decene dimers made using BF3,
and decene
dimers from either BF3 or aluminum halide catalysis, gave the
same peaks (but not the same relative peak heights) in standard
proton decoupled 13C NMR spectra. We therefore expect that BF3
and aluminum halides oligomerize 1-decene by common
mechanisms. Reactions were batch (not continuous) and allowed
us to sample reaction mixtures at a number of reaction times
and levels of monomer conversion.
Table 1. Double bond positions in residual decenes
Decene isomer
AlX3 catalysis
BF3 catalysis
Reactions with 1-decene were instructive. Figure 2 shows
chromatograms of the C20 olefin fractions at 47% and 75%
conversion during a representative 1-decene oligomerization. A
large number of C20 isomers are detectable in both samples,
and both chromatograms appear to show similar peaks for the
C20 products. At 47% conversion, 93% of the unconverted
1-decene
2-decenes
3-decenes
4-decenes
5-decenes
69.0%
15.0%
6.1%
7.0%
2.9%
76.3%
9.4%
5.4%
5.4%
3.4%
wileyonlinelibrary.com/journal/poc
Copyright © 2012 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2012, 25 1409–1417

PROTONATED CYCLOPROPYL INTERMEDIATES IN POLYALPHAOLEFIN SYNTHESIS
mole % 1-tetradecene. When 39% of the C14 monomer had
converted to oligomers, only 24% of the C10 monomer had
converted to oligomers. Extensive skeletal isomerization in the
dimer fraction does not require extensive double bond isomeri-
zation in the monomer.
comparable to that of internal decenes, but the C12 isomers un-
derwent no observed skeletal isomerization. Figure 5 shows the
gas chromatograms for the C12 isomers before and after treat-
ment with AlBr3 for several hours. These observations argue that
Structure I in Figs. 1 and 3 does not undergo skeletal isomeriza-
tion during PAO synthesis. Figure 6 summarizes our findings re-
garding which types of carbocations do not appear to skeletally
isomerize in PAO synthesis.
We propose that 2-decyl carbocation combines with 1-decene
to form a protonated cyclopropyl intermediate directly and that
Structure I in Figs. 1 and 3 is not an intermediate in the formation
of the protonated cyclopropyl intermediate. At least one previ-
ous report has suggested a similar interaction between olefins
and carbocations,[13] but we are unaware of previous reports of
rapid, multiple skeletal isomerizations from such intermediates.
Examples of cyclopropanation reactions are known[14] but do
not appear to occur by direct interaction of a carbocation with
an olefin.
Skeletal rearrangements evidently do not occur in the mono-
mer, but some type of rapid isomerization clearly occurs during
formation of dimer. There must be a rearrangement mechanism
open to the dimer that is not open to the monomer. The original
Shubkin mechanism (Fig. 3) proposes a protonated cyclopropyl
intermediate that can open to a tertiary carbenium ion. Rear-
rangement to this tertiary carbenium ion may be favorable, but
one such isomerization does not immediately explain the steps
involved in producing fifty or more C20 isomers, as there
appears to be no clear reason why the tertiary carbenium ion
in Fig. 3 would continue to isomerize to other carbenium ions
faster than it would either eliminate a proton to form olefin or re-
act with 1-decene to make trimer. Moreover, if this tertiary carbe-
nium ion did react with 1-decene to make trimer, the trimer
would contain a quaternary carbon, which is not observed in
the product mixtures.[5]
We subjected 2-octyl-1-decene to oligomerization conditions
with AlBr3. This olefin should form a tertiary carbocation immedi-
ately upon contact with catalyst. While we observed double
bond isomerization and oligomerization in this reaction mixture,
the monomer showed no evidence of skeletal isomerization. For-
mation of a tertiary carbenium ion evidently did not immediately
lead to extensive skeletal isomerization in the monomer.
To model Structure I in Figs. 1 and 3, we prepared a mixture
containing about 29 wt % 7-methyl-4-undecenes, 36 wt %
7-methyl-5-undecenes, 15 wt % 4-dodecenes, and 15 wt %
5-dodecenes, with 5% other C12 isomers. Under oligomerization
conditions, half the 7-methyl-5-undecenes and half the
7-methyl-4-undecenes (~32% of total monomer) should immedi-
ately generate carbocations with methyl groups two carbons
away from the centers of positive charge (Fig. 4). If Structure I
represents the first intermediate along the route to skeletal rear-
rangement in the PAO dimer fractions, then this C12 monomer
should undergo skeletal rearrangement under oligomerization
conditions. It did not. In a competing oligomerization with linear
internal decenes, these C12 isomers formed oligomers at a rate
Figure 4. Comparison of Shubkin’s proposed C20 intermediate with C12
carbocation expected to form from methyl undecenes
Figure 3. Shubkin mechanism showing protonated cyclopropyl
Figure 5. Gas chromatograms of C10 and C12 olefins before and after
intermediate
treatment with AlBr3 for oligomerization
J. Phys. Org. Chem. 2012, 25 1409–1417
Copyright © 2012 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/poc

J. C. GEE, B. L. SMALL AND K. D. HOPE
1. Carbon 8 binds with carbon 11
2. Carbon 8 binds with carbon 12
3. Carbon 10 binds with carbon 11
4. Carbon 10 binds with carbon 12
5. Carbon 13 binds with carbon 9
6. Carbon 13 binds with carbon 11
Each of these options appears to be a concerted step akin to
an SN2 reaction, in which the alpha carbon acts as a nucleophile
and a protonated corner carbon acts as a leaving group.
Figure 6. Carbocations that do not skeletally isomerize during PAO
synthesis
Figure 7 shows our proposed mechanism for the origin of
skeletal isomerization in PAO synthesis. A protonated cyclopro-
pyl intermediate forms and then begins rapid isomerizations.
These isomerizations may not all be equally fast: Route 1 above
may be faster than route 2, for example, if binding with a
secondary ring carbon is faster than binding with a tertiary ring
carbon. In fact, constraints of the zeolitic cage in the SAPO cata-
lyst prevent isomerizations that would result in alkyl branches
larger than a methyl group, so all isomerization steps on the
SAPO catalyst involve a secondary ring carbon. We assume this
constraint does not apply to the current homogeneous reaction.
(Following this mechanism, internal decene monomers cannot
form protonated cyclopropyl intermediates having methylene
carbons in the rings, but internal olefins still show extensive
skeletal isomerization at the dimer stage.)
The intermediates in Fig. 7 can then either isomerize again,
react directly with monomer to make trimer, or they can open
to form carbenium ions that eliminate a proton to form C20 ole-
fins. Rings that open have three different bonds that can break,
so on first thought, each cyclopropyl intermediate appears capa-
ble of forming up to three different skeletal isomers. However,
our previously reported SAPO observations[11] suggest that some
ring opening mechanisms are much more likely than others.
Figure 8 shows a typical protonated cyclopropyl intermediate
expected to form during the skeletal isomerization of a linear
olefin on a SAPO or other solid acid catalyst. We think the
There is precedent to support the proposal that a protonated
cyclopropyl intermediate can undergo rapid skeletal isomeriza-
tion and thus lead to the multiple skeletal isomers observed in
PAO synthesis. We recently reported observations on the double
bond and skeletal isomerization of linear olefins on a SAPO cat-
alyst.[11] On this catalyst, double bonds migrated in a step-wise
manner. Additionally, skeletal isomerization to methyl substi-
tuted olefins occurred, presumably through a protonated cyclo-
propyl intermediate.[15–20] During skeletal isomerization, the
methyl groups formed at nearly random positions along the car-
bon chain, regardless of the double bond location in the starting
olefin, and the catalyst was inactive towards linear alkanes under
the conditions of our experiments. We concluded that the
intermediate protonated cyclopropyl intermediate isomerized
extremely rapidly (and repeatedly), essentially to an equilibrium
distribution of cyclic isomers, before it opened to form the
methyl substituted product. If this protonated cyclopropyl
intermediate isomerizes extremely rapidly, then perhaps the
protonated cyclopropyl intermediate proposed by Shubkin in
PAO synthesis isomerizes in a similar fashion. If it does, then
we can predict what structures should form from such a
mechanism.
Figure 3 shows the initial steps of Shubkin’s proposed mecha-
nism and the first protonated cyclopropyl intermediate
(Structure II). Our data in Fig. 5 demonstrate that the mechanism
shown in Fig. 3 is not entirely correct: Structure I in Fig. 3
apparently does not lead to skeletal isomerization in the
dimer. However, protonated cyclopropyl intermediates are
broadly accepted as intermediates during skeletal isomerization,
and we propose that Structure II in Fig. 3 forms directly from
combination of 2-decyl carbocation and 1-decene. Previous
observations regarding skeletal isomerization of linear olefins[11]
suggest that carbons alpha to the ring in Structure II/Fig. 3, such
as carbons 8, 10, and 13, can form new cyclopropyl rings by
binding with either of the two carbons on the opposite side of
the cyclopropyl ring. This type of binding is consistent with the
observations long reported for the interactions of a cyclopro-
pane ring with an alpha carbon,[21–27] and isotopic scrambling
experiments[28,29] show that such interchanges are rapid for
a sec-butyl cation. Indeed, there are numerous reports of
rapid migrations involving three-membered rings within cyclic
systems,[30–35] and the skeletal isomerization route we are
proposing for PAO may simply be an extension of these transfor-
mations to intermediates that do not have a large ring system.
The well-known investigations of the 2-norbornyl cation[36–40]
also demonstrate the stabilizing effects of corner-protonated
cyclopropyl species.
Structure II/Fig. 3 therefore has six different modes for ex-
tremely rapid isomerization to a new cyclopropyl intermediate:
Figure 7. Proposed rapid isomerization of protonated cyclopropyl
intermediates
wileyonlinelibrary.com/journal/poc
Copyright © 2012 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2012, 25 1409–1417

PROTONATED CYCLOPROPYL INTERMEDIATES IN POLYALPHAOLEFIN SYNTHESIS
Figure 8. Ring opening mechanism for skeletal isomerization of linear
olefins observed on SAPO catalyst (Tertiary carbenium ion highly preferred)
intermediates in our previous experiments opened by 1,2
hydride shifts that led always (or almost always) to a tertiary
rather than a secondary carbenium ion. The acidic proton
associated with such cyclopropyl intermediates appears to pro-
tonate either edge-wise[16,18] on one of the carbon–carbon
bonds or on a corner carbon[19,20] of the ring and is rapidly
mobile. Calculations have indicated that corner-protonated
species are slightly more stable than edge-protonated
species and that proton migration occurs rapidly through the
edge-protonated intermediate.[41] The carbon–carbon bond that
breaks upon ring opening is likely one of the carbon–carbon
bonds associated with a protonated corner carbon. If the rapidly
migrating rings opened to secondary carbenium ions, then the
ring isomerizations could serve as a route to rapid, non-stepwise
double bond isomerization, which we did not observe in
those experiments. The protonated cyclopropyl intermediates
appeared to open only to the tertiary carbenium ions shown
in Fig. 8; opening to the secondary carbenium ion apparently
did not occur. We therefore think that ring openings for
intermediates like those in Fig. 7 should follow routes like those
shown in Fig. 9.
Figure 9. Expected ring opening mechanisms allowed during PAO
synthesis
We think trimer likely forms by direct reaction between
1-decene and a protonated cyclopropyl intermediate. Our pro-
posed mechanism requires that the protonated cyclopropyl
intermediates open to tertiary carbenium ions. If 1-decene
reacted with a tertiary carbenium ion in a propagation step, a
quaternary carbon would result in the product, and NMR
analyses have not revealed quaternary carbons in PAO samples.[5]
Direct reaction of 1-decene with a protonated cyclopropyl inter-
mediate, however, permits the formation of trimer molecules
without quaternary carbons (See Fig. 10).
Figure 10. Direct reaction of protonated cyclopropyl intermediate with
monomer
Explaining the mechanism of propagation beyond the trimer
stage is problematic. If propagation continued via the simple cat-
ionic mechanism, there would be no further skeletal isomeriza-
tion in the growing chain, consistent with actual NMR data
reported for samples of trimer and tetramer.[1] This observation
argues that carbocations at the trimer stage do not combine
with monomer directly to make protonated cyclopropyl inter-
mediates that introduce additional methyl groups. We have no
satisfactory explanation for this behavior, except to suggest that
the branched structures formed at the dimerization stage steri-
cally inhibit direct formation of another protonated cyclopropyl
intermediate.
We constructed a computer model to predict the structures of
C20 isomers expected after a number of repeated isomerizations
of Structure II in Fig. 3. The model started with 3000[42] inter-
mediates like Structure II/Fig. 3. For each “molecule,” the com-
puter selected a random, allowed mode of isomerization and
Table 2. Simulation results for random isomerizations of
Structure II/Fig. 3
# Isomerizations
# C20 Isomers
% as Vicinal dimethyl
1
2
3
4
11
23
44
63
85
189
368
463
36.3
33.6
30.9
27.6
24.9
14.6
6.4
5
10
25
50
3.6
J. Phys. Org. Chem. 2012, 25 1409–1417
Copyright © 2012 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/poc

J. C. GEE, B. L. SMALL AND K. D. HOPE
replaced the starting structure with the resulting new one.
(In this approach, all isomerizations were equally probable,
which may not reflect reality.) We could repeat this process any
desired number of times, thus simulating different rates of
isomerization. At the end of all the isomerizations, the computer
selected a random, allowed mode of ring opening for each
cyclopropyl intermediate. The resulting carbenium ion elimi-
nated a proton to give an olefin that, upon hydrogenation, gave
the observed C20 paraffin.
Table 2 shows the number of skeletal isomers predicted for
a number of different random isomerizations per cyclopropyl
intermediate. The number of isomers listed does not include
stereoisomers. GC/MS data have been consistent with the
presence of resolvable diasteromers,[6] and we could add R
and S options for each asymmetric carbon present in these
structures, thus increasing the actual number of unique struc-
tures that one might detect in a detailed chromatographic
analysis.
Table 3. Product distribution predicted for three random isomerizations of Structure II/Figure 3
One possible route for new C–C bonds
at each ring isomerization
Rank
Isomer type
Main chain
%
Running total
Step 1
Step 2
Step 3
1
2
3
4
5
6
7
8
9,10-Dimethyl
8,9-Dimethyl
9-Isopropyl
8,10-Dimethyl
8-Ethyl-9-Methyl
8-(2-Methylpropyl)
10-Methyl
9-Ethyl
9-Methyl
9-Ethyl-8-Methyl
8-Isopropyl
7,10-Dimethyl
7-Ethyl-9-Methyl
7,9-Dimethyl
8-Methyl-7-Propyl
7-(3-Pentyl)
7-Ethyl-8-Methyl
7-(2-Butyl)
7-Methyl-8-Propyl
6-Ethyl-9-Methyl
9-Propyl
6,10-Dimethyl
8-Ethyl-7-Methyl
8-(3-Methylbutyl)
7-Isopropyl
7,8-Dimethyl
9-Ethyl-7-Methyl
8-Methyl-6-Propyl
7,8-Diethyl
6-Ethyl-8-Methyl
8-Ethyl
6-Butyl-7-Methyl
7-Butyl -6-Methyl
6-(3-Methylbutyl)
8-Propyl
Octadecane
Octadecane
Heptadecane
Octadecane
Heptadecane
Hexadecane
Nonadecane
Octadecane
Nonadecane
Heptadecane
Heptadecane
Octadecane
Heptadecane
Octadecane
Hexadecane
Pentadecane
Heptadecane
Hexadecane
Hexadecane
Heptadecane
Heptadecane
Octadecane
Heptadecane
Pentadecane
Heptadecane
Octadecane
Heptadecane
Hexadecane
Hexadecane
Heptadecane
Octadecane
Pentadecane
Pentadecane
Pentadecane
Heptadecane
Octadecane
Hexadecane
Hexadecane
Octadecane
Heptadecane
Hexadecane
Tetradecane
Hexadecane
Hexadecane
Tetradecane
21.7
8.8
8.5
7.3
6.0
5.8
5.6
5.4
5.0
4.7
4.5
2.4
2.1
1.5
1.1
1.0
1.0
0.8
0.8
0.7
0.7
0.6
0.5
0.4
0.3
0.3
0.3
0.2
0.2
0.2
0.2
0.2
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.0
0.0
0.0
0.0
0.0
21.7
30.5
39.0
46.3
52.3
58.1
63.8
69.2
74.2
78.9
83.4
85.8
87.9
89.4
90.5
91.5
92.5
93.4
94.2
94.9
95.6
96.1
96.6
97.0
97.4
97.7
97.9
98.2
98.4
98.6
98.8
99.0
99.1
99.2
99.4
99.5
99.6
99.7
99.8
99.9
99.9
99.9
100.0
100.0
100.0
13 to 11
12 to 8
11 to 10
12 to 10
13 to 9
12 to 8
12 to 8
13 to 9
11 to 8
13 to 9
12 to 10
13 to 11
13 to 9
12 to 8
13 to 11
13 to 9
11 to 8
11 to 10
12 to 10
13 to 11
13 to 11
13 to 11
11 to 8
12 to 8
11 to 8
11 to 8
13 to 9
13 to 11
13 to 9
11 to 8
11 to 8
13 to 9
13 to 9
12 to 8
11 to 8
11 to 8
12 to 8
13 to 9
12 to 8
12 to 8
11 to 8
13 to 9
13 to 9
13 to 9
13 to 9
11 to 9
10 to 8
12 to 10
12 to 11
12 to 10
13 to 9
11 to 8
12 to 10
12 to 11
13 to 10
11 to 10
14 to 12
14 to 9
11 to 8
14 to 12
13 to 10
9 to 7
10 to 8
13 to 9
14 to 11
13 to 9
14 to 11
11 to 7
12 to 7
12 to 8
12 to 8
13 to 10
14 to 12
13 to 10
9 to 7
12 to 10
9 to 7
12 to 11
13 to 11
13 to 12
14 to 12
11 to 10
11 to 9
11 to 10
14 to 9
10 to 8
15 to 13
15 to 13
11 to 7
14 to 9
14 to 9
9 to 6
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
9 to 7
14 to 9
15 to 13
13 to 10
15 to 11
9 to 7
13 to 7
9 to 7
9 to 7
14 to 10
15 to 12
14 to 9
8 to 6
10 to 7
15 to 9
15to 9
12 to 6
10 to 7
11 to 6
8 to 6
14 to 10
9 to 6
9 to 7
14 to 9
14 to 9
12 to 7
9 to 7
11 to 7
12 to 7
14 to 9
9 to 7
9 to 7
9 to 7
14 to 9
14 to 9
14 to 9
14 to 9
6,9-Dimethyl
6-(2-Methylpropyl)
8-Butyl
6,7-Dimethyl
6-Isopropyl
6-Methyl-7-Propyl
5-Methyl-6-Pentyl
9-Methyl -8-Propyl
8-Methyl-9-Propyl
6-(2-Hexyl)
9 to 6
9 to 6
15 to 9
14 to 8
13 to 8
15 to 9
wileyonlinelibrary.com/journal/poc
Copyright © 2012 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2012, 25 1409–1417

PROTONATED CYCLOPROPYL INTERMEDIATES IN POLYALPHAOLEFIN SYNTHESIS
Table 2 also shows the percent of products that were vicinal
dimethyl octadecanes. Previous investigators have reported that
up to about 30% of the C20 products could be vicinal dimethyl
octadecanes.[6] If that report be correct, then about three isomer-
izations per intermediate would be enough to account for the
level of vicinal dimethyl octadecanes and the presence of more
than 40 skeletal isomers. In our SAPO work on skeletal isomeriza-
tion of linear olefins, the rings appeared to isomerize fast enough
to yield equal populations of nine different cyclic intermediates
before opening to a methyl substituted chain: We observed the
isomerization of 11-dococene to nearly equal quantities of 2, 3,
4, and 5 + -methyl substituted chains. Our kinetic calculations
have shown that the rate of ring isomerization must have been
at least 500 times faster than the rate of ring opening to account
for this observation.
Table 3 shows a typical product distribution predicted from a
simulation that used three random isomerizations for each
Structure II in Fig. 3. The model assumes that all the cyclic inter-
mediates open equally fast to form C20 isomers and that they all
react equally fast with 1-decene to make trimer. The distribution
has many products previously suggested to be present in PAO 2.
Noteworthy is the prediction of 9,10-dimethyl octadecane as the
product of highest concentration, which previous work has
suggested is the structure present in highest quantity.[5,6]
Additionally, this product mixture shows an average of 3.7
methyl groups per chain, consistent with NMR data on real
samples.[1–5] We note, however, that this distribution contains
some structures having isopropyl groups, which previous NMR
analyses have failed to detect.[5] The evidence for there being
no isopropyl isomers in PAO 2 appears to be the absence of a
predicted methyl peak in the 13C NMR spectra of PAO 2
samples. This predicted methyl peak might be overwhelmed
by nearby methylene peaks, so the evidence against isopropyl
isomers might not be conclusive. If isopropyl isomers are not
actually present in PAO 2, some isomerizations we propose
may be less likely than others. For example, formation of much
isopropyl substituted product would require that carbon 8 in
Structure II/Fig. 3 bind with carbon 12 at a similar rate as binding
at carbon 11, which might very well be unlikely.
In the event of 1-decene isomerization to 2-decenes, we have
the possibility that a 3-decyl carbenium ion reacts with 1-decene
to make the cyclic intermediate having an ethyl rather than a
methyl substituent on carbon 9 in Structure II/Fig. 3. Simulations
showed that products derived from this intermediate were es-
sentially the same as those derived from Structure II, although
the predicted distribution of those products was not the same.
Double bond isomerization in the monomer therefore appears
to affect the possible distribution of dimeric products but not
the list of available structures. As an illustration, Fig. 11 shows
an example mechanistic route to 7-butyl-6-methyl pentadecane,
a structure difficult to explain by other mechanisms.
Table 4 lists products that previous workers have identified in
PAO 2.[5,6] We can account for the formation of every one of
them using the current mechanism operating on Structure II/
Table 4. Products previously identified in PAO 2 and pre-
dicted by proposed mechanism
Reference #
2,3-dimethyl
3,4-dimethyl
4,5-dimethyl
5,6-dimethyl
6,7-dibutyl
Octadecane
Octadecane
Octadecane
Octadecane
Dodecane
6
6
6
6
6
6,7-dimethyl
6-butyl
Octadecane
Hexadecane
Tetradecane
Tridecane
Tridecane
Dodecane
6
6
6
6
6
6
6
5
6
6
5
6
5
5
6
6
6-butyl-7-ethyl
6-butyl-7-propyl
6-ethyl-7-pentyl
6-pentyl-7-propyl
7-methyl -6-propyl
7,10-dimethyl
7,8-dimethyl
7,8-dipropyl
7-butyl-8-methyl
7-propyl
8,10-dimethyl
8,9-diethyl
8,9-dimethyl
8-ethyl
9-ethyl -8-methyl
8-methyl-9-propyl
9,10-dimethyl
9-methyl
Hexadecane
Octadecane
Octadecane
Tetradecane
Pentadecane
Heptadecane
Octadecane
Hexadecane
Octadecane
Octadecane
Heptadecane
Hexadecane
Octadecane
Nonadecane
Hexadecanes
5
5
5,6
5,6
6
butyl (5,7, and 8)
Figure 11. Example mechanistic route to 7-butyl-6-methyl pentadecane
J. Phys. Org. Chem. 2012, 25 1409–1417
Copyright © 2012 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/poc

J. C. GEE, B. L. SMALL AND K. D. HOPE
Table 5. Previously identified octane isomers from butene dimerizations
Products from dimerization of
1-butene and cis-2-butene[2]
Identified in
mixture
Quaternary
carbon
Predicted by simple mechanism
Predicted by new
proposed mechanism
if isobutene forms
2,2,4-trimethyl pentane
2,2-dimethyl hexane
2,2,3-trimethyl pentane
3,3-dimethyl hexane
2,3,3-trimethyl pentane
3-methyl-3-ethyl pentane
3,4-dimethyl hexane
2,3-dimethyl hexane
2,4-dimethyl hexane
2-methyl-3-ethyl pentane
3-methyl heptanes
4-methyl heptanes
2,5-dimethyl hexane
2-methyl heptanes
2,3,4-trimethyl pentane
3-ethyl hexane
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Fig. 3. The minimum number of random isomerizations required
to obtain a distribution that includes all the products listed in
Table 4 results in a mixture that has substantially less than 30%
vicinal dimethyl products, suggesting that some isomerizations
are probably faster than others or that some dimer intermediates
react with 1-decene faster than others. Alternatively, low levels
of some of these products could result from reactions involving
internal decenes.
Acknowledgements
We thank Dr. Robert Coffin and Dr. Steve Herron for helpful
discussions, Eric Fernandez for the preparation of the dodecene
mixture, and Linda Nemcheck for conducting some of the
oligomerization experiments. We also thank Chevron Phillips
Chemical Company for permission to publish this work.
Onopchenko, et al. reported a list of 15 products they
identified in the octane fractions isolated after dimerization of
1-butene and of cis-2-butene.[2] Their catalyst was a complex of
BF3 and mannitol, because they were unable to isolate dimer
fractions when they used other BF3 catalysts. They identified
six products having quaternary carbons, which they thought
had originated after skeletal isomerization of monomer to
isobutene. Of the nine remaining products, our proposed
mechanism predicts eight of them, and one product predicted
by our mechanism is not listed among Onopchenko’s products.
Our model predicts five identified products that appear to have
no other explainable origin (Table 5). The new mechanism
predicts that 1-butene and 2-butene should give rise to the same
products but not in the same distributions, consistent with
Onopchenko’s reported results.
REFERENCES
[1] R. L. Shubkin, M. S. Baylerian, A. R. Maler, Ind. Eng. Chem. Prod. Res.
Dev. 1980, 19, 15.
[2] A. Onopchenko, B. L. Cupples, A. N. Kresge, Ind. Eng. Chem. Prod. Res.
Dev. 1983, 22, 182.
[3] G. L. Driscoll, S. J. G. Linkletter, Synthesis of Synthetic Hydrocarbons
via Alpha Olefins, AFWAL-TR-85-4066; NTIS: 1985, 96 pp.
[4] R. L. Shubkin, M. E. Kerkemeyer, J. Syn. Lub. 1991, 8, 115.
[5] G. S. Kapur, A. S. Sarpal, R. Sari, S. K. Jain, S. P. Srivastava A. K.
Bhatnagar, J. Syn. Lub. 1998, 15, 177.
[6] S. S. Scheuermann, S. Eibl, P. Bartl, Lubrication Science 2011, 23, 221.
[7] B. L. Cupples, W. J. Heilman, A. N. Kresge, Method of oligomerizing
1-olefins. US Patent 4,045,507, 1977.
[8] J. A. Brennan. Polymerization of olefins with BF3. US Patent 3,382,291,
1968.
[9] F. C. Whitmore, Ind. Eng. Chem. 1934, 26, 94.
[10] B. L. Small, A. J. Marcucci, Organometallics 2001, 20, 5738.
[11] J. C. Gee, D. S. Prampin, Appl Cat A: Gen 2009, 360, 71.
[12] J. C. Gee, D. S. Prampin, Anal. Chem. 2009, 81, 1646.
[13] C.-H. R. King, C. D. Poulter, J. Am. Chem. Soc. 1982, 104, 1413.
[14] M. Sugawara, J. Yoshida, J. Am. Chem. Soc. 1997, 119, 11986.
[15] G. M. Kramer, J. Am. Chem. Soc. 1969, 91, 4819.
[16] C. J. Collins, Chem. Rev. 1969, 69, 543.
CONCLUSIONS
The structures predicted by this model appear to be in accord
with the types of structures suggested previously for PAO 2,
and the model predicts products previously identified in
mixtures of butene dimers and decene dimers. Three repeated
isomerizations of the initial protonated cyclopropyl intermediate
are enough to generate enough isomers to account for the
variety of structures present in the C20 fraction isolated from
commercial PAO processes. This rapid isomerization of proton-
ated cyclopropyl intermediates, previously reported for skeletal
isomerization of linear olefins, is a possible explanation for the
large number of skeletal isomers observed in commercial PAO.
This intermediate appears to form directly from combination of
2-decyl carbocation with 1-decene.
[17] D. Fărcaşiu, S. H. Norton, D. Hâncu, J. Am. Chem. Soc. 2000,
122, 668.
[18] C. C. Lee, S. Vassie, E. C. F. Ko, J. Am. Chem. Soc. 1972, 94, 8931.
[19] C. H. DePuy, A. H. Andrist, P. C. Fünfschilling, J. Am. Chem. Soc. 1974,
96, 948.
[20] M. J. S. Dewar, E. F. Healy, J. M. Ruiz, J. Chem. Soc., Chem. Comm.
1987, 12, 943.
[21] P. R. Schleyer, G. W. Van Dine, J. Am. Chem. Soc. 1966, 88, 2321.
[22] J. B. Lambert, A. P. Jovanovich, J. W. Hamersma, F. R. Koeng, S. S.
Oliver, J. Am. Chem. Soc. 1973, 95, 1570.
[23] P. J. Chenier, T. M. Jenson, W. D. Wulff, J. Org. Chem. 1982,
47, 770.
[24] G. D. Sargent, N. Lowry, S. D. Reich, J. Am. Chem. Soc. 1967, 89, 5985.
[25] J. D. Roberts, R. H. Mazur, J. Am. Chem. Soc. 1951, 73, 2509.
wileyonlinelibrary.com/journal/poc
Copyright © 2012 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2012, 25 1409–1417

PROTONATED CYCLOPROPYL INTERMEDIATES IN POLYALPHAOLEFIN SYNTHESIS
[26] H. Hart, P. A. Law, J. Am. Chem. Soc. 1964, 86, 1957.
[27] H. Hart, P. A. Law, J. Am. Chem. Soc. 1962, 84, 2462.
[28] G. E. Walker, O. Kronja, M. Saunders, J. Org. Chem. 2004, 69, 3598.
[29] M. Saunders, E. L. Hagen, J. Rosenfeld, J. Am. Chem. Soc. 1968, 90, 6882.
[30] R. F. Childs, C. V. Rogerson, J. Am. Chem. Soc. 1976, 98, 6391.
[31] R. F. Childs, S. Winstein, J. Am. Chem. Soc. 1974, 96, 6409.
[32] P. Vogel, M. Saunders, N. M. Hasty, J. A. Berson, J. Am. Chem. Soc.
1971, 93, 1551.
[36] G. A. Olah, A. M. White, J. R. DeMember, A. Commeyras, C. Y. Lui,
J. Am. Chem. Soc. 1970, 92, 4627.
[37] C. S. Yannoni, V. Macho, P. C. Myhre, J. Am. Chem. Soc. 1982, 104, 7380.
[38] H. C. Brown, M. Periasamy, D. P. Kelly, J. J. Giansiracusa, J. Org. Chem.
1982, 47, 2089.
[39] G. A. Olah, G. K. S. Prakash, D. G. Farnum, T. P. Clausen, J. Org. Chem.
1983, 48, 2146.
[40] M. Saunders, M. R. Kates, J. Am. Chem. Soc. 1983, 105, 3571.
[41] W. Koch, B. Liu, J. Am. Chem. Soc. 1989, 111, 3479.
[42] Model predictions were statistically consistent, as long as the
number of molecules was at least 2000 nd.
[33] R. F. Childs, S. Winstein, J. Am. Chem. Soc. 1968, 90, 7146.
[34] D. W. Swatton, H. Hart, J. Am. Chem. Soc. 1967, 89, 5075.
[35] W. Kirmse, J. Streu, J. Org. Chem. 1987, 52, 515.
J. Phys. Org. Chem. 2012, 25 1409–1417
Copyright © 2012 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/poc